BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Australia/Melbourne
X-LIC-LOCATION:Australia/Melbourne
BEGIN:DAYLIGHT
TZOFFSETFROM:+1000
TZOFFSETTO:+1100
TZNAME:AEDT
DTSTART:19721003T020000
RRULE:FREQ=YEARLY;BYMONTH=4;BYDAY=1SU
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:19721003T020000
TZOFFSETFROM:+1100
TZOFFSETTO:+1000
TZNAME:AEST
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260114T163641Z
LOCATION:Meeting Room C4.11\, Level 4 (Convention Centre)
DTSTART;TZID=Australia/Melbourne:20231215T102500
DTEND;TZID=Australia/Melbourne:20231215T104000
UID:siggraphasia_SIGGRAPH Asia 2023_sess135_papers_248@linklings.com
SUMMARY:Text-Guided Synthesis of Eulerian Cinemagraphs
DESCRIPTION:Aniruddha Mahapatra (Carnegie Mellon University); Aliaksandr S
 iarohin, Hsin-Ying Lee, and Sergey Tulyakov (Snap Inc.); and Jun-Yan Zhu (
 Carnegie Mellon University)\n\nWe introduce Text2Cinemagraph,  a fully aut
 omated method for creating cinemagraphs from text descriptions - an especi
 ally challenging task when prompts feature imaginary elements and artistic
  styles, given the complexity of interpreting the semantics and motions of
  these images. We focus on cinemagraphs of fluid elements, such as flowing
  rivers, and drifting clouds, which exhibit continuous motion and repetiti
 ve textures. Existing single-image animation methods fall short on artisti
 c inputs, and recent text-based video methods frequently introduce tempora
 l inconsistencies, struggling to keep certain regions static. To address t
 hese challenges, we propose an idea of synthesizing image twins from a sin
 gle text prompt - a pair of an artistic image and its pixel-aligned corres
 ponding natural-looking twin. While the artistic image depicts the style a
 nd appearance detailed in our text prompt, the realistic counterpart great
 ly simplifies layout and motion analysis.  Leveraging existing natural ima
 ge and video datasets, we can accurately segment the realistic image and p
 redict plausible motion given the semantic information. The predicted moti
 on can then be transferred to the artistic image to create the final cinem
 agraph. Our method outperforms existing approaches in creating cinemagraph
 s for natural landscapes as well as artistic and other-worldly scenes, as 
 validated by automated metrics and user studies. Finally, we demonstrate t
 wo extensions: animating existing paintings and controlling motion directi
 ons using text.\n\nRegistration Category: Full Access\n\nSession Chair: Ch
 ongyang Ma (ByteDance)\n\n
URL:https://asia.siggraph.org/2023/full-program?id=papers_248&sess=sess135
END:VEVENT
END:VCALENDAR